Configuration

Aripa is configured with environment variables and a config.json file. The easiest way to create or update config.json is:

bun run onboard

Environment Variables

Variable	Description
`TOKEN`	Discord bot token. Required to run the bot.
`PREFIX`	Text command prefix. Defaults to `-`.
`LOG_LEVEL`	Logging level. Defaults to `info`.
`DATABASE_PATH`	SQLite database path. Defaults to `aripa.sqlite`.
`CONFIG_PATH`	Runtime config path. Defaults to `config.json`.
`OPENAI_API_KEY`	Used by OpenAI models unless a provider-specific `apiKeyEnv` is configured.
`OPENROUTER_API_KEY`	Used by OpenRouter models unless `providers.openrouter.apiKeyEnv` is configured.
`AI_GATEWAY_API_KEY`	Used by Vercel AI Gateway unless `providers.gateway.apiKeyEnv` is configured.
`GOOGLE_GENERATIVE_AI_API_KEY`	Used by Gemini web search unless `providers.google.apiKeyEnv` is configured.
`ARIPA_UPDATE_GITHUB_REPO`	Temporarily overrides `updates.githubRepo` for `bun run update`.
`GITHUB_TOKEN`	Optional token for GitHub release API requests during updates.
`ARIPA_RELEASE_PUBLIC_KEY_PEM`	Release verification public key for custom update sources.
`ARIPA_RELEASE_PUBLIC_KEY_PEM_B64`	Base64 release verification public key for custom update sources.

config.json

Example:

{
  "name": "Aripa",
  "stylePrompt": "match",
  "agentRateLimitMessagesPerMinute": 10,
  "agentTimeoutMs": 60000,
  "agentMaxConcurrentRequests": 4,
  "agentMaxConcurrentRequestsPerGuild": 2,
  "logPrivacy": true,
  "models": {
    "agent": {
      "provider": "openai",
      "model": "gpt-5.5",
      "reasoningEffort": "low"
    },
    "summarizer": {
      "provider": "openai",
      "model": "gpt-5.4-nano",
      "reasoningEffort": "low"
    },
    "web": {
      "enabled": true,
      "provider": "google",
      "model": "gemini-2.5-flash"
    }
  },
  "providers": {},
  "updates": {
    "enabled": true,
    "githubRepo": "aripabot/aripa"
  },
  "allowlistedServerIds": [
    "123456789012345678"
  ]
}

Runtime Fields

Field	Description
`name`	Bot instance name used by runtime prompts and display text.
`stylePrompt`	Agent response style, such as `match`, `friendly`, `playful`, `concise`, or `formal`.
`allowlistedServerIds`	Discord server IDs where this bot is allowed to operate. At least one server ID is required when using onboarding.
`agentRateLimitMessagesPerMinute`	Per-user agent mention limit. Use `null` to disable.
`agentTimeoutMs`	Maximum time, in milliseconds, for an agent response before timing out. Defaults to `60000`.
`agentMaxConcurrentRequests`	Global maximum number of concurrent agent requests. Defaults to `4`.
`agentMaxConcurrentRequestsPerGuild`	Maximum number of concurrent agent requests per Discord server. Defaults to `2`.
`logPrivacy`	Redacts agent tool inputs and outputs from logs when `true`. Defaults to `true`.
`models.agent`	Model used for agent responses.
`models.summarizer`	Model used to summarize long context for agent tooling.
`models.web`	Gemini web-search model configuration. Set `enabled` to `false` to disable web search.
`providers`	Optional provider overrides such as `baseURL` and `apiKeyEnv`.
`updates`	Release updater configuration used by `bun run update`.

Models

models.agent and models.summarizer support these providers:

openai
openrouter
gateway
ollama
lmstudio

models.web always uses Google Gemini. The provider value is normalized to google.

Reasoning effort may be one of:

minimal
low
medium
high

Reasoning effort is currently applied to OpenAI model calls.

Provider Overrides

Use providers to customize API keys or base URLs.

{
  "providers": {
    "openrouter": {
      "apiKeyEnv": "MY_OPENROUTER_KEY"
    },
    "ollama": {
      "baseURL": "http://localhost:11434/v1"
    },
    "lmstudio": {
      "baseURL": "http://localhost:1234/v1"
    },
    "google": {
      "apiKeyEnv": "MY_GEMINI_KEY"
    }
  }
}

Each provider supports:

Field	Description
`baseURL`	Optional custom API base URL. Useful for local or compatible providers.
`apiKeyEnv`	Optional environment variable name to read the API key from.

Unknown provider keys are ignored.

Updates

The updater is configured with the updates block:

{
  "updates": {
    "enabled": true,
    "githubRepo": "aripabot/aripa"
  }
}

Field	Description
`enabled`	Enables or disables `bun run update`. Defaults to `true`.
`githubRepo`	GitHub repository to read published releases from, in `owner/repo` format. Defaults to `aripabot/aripa`.
`releasePublicKeyPem`	Optional PEM public key used to verify releases from a custom update source.
`releasePublicKeyPemBase64`	Optional base64-encoded PEM public key used to verify releases from a custom update source.

Official Aripa releases use the built-in official release verification key. Forks or custom release repositories should configure their own public key or set ARIPA_RELEASE_PUBLIC_KEY_PEM_B64.

Privacy

Keep logPrivacy set to true when logs leave the host or may be viewed by people who should not see full agent tool inputs and outputs.